about me
I am a PhD candidate in the Earth System Science department at the Stanford Doerr School of Sustainability. I am part of the ECHO Lab and lucky to be advised by Marshall Burke. I am supported by a Stanford Data Science Fellowship and the Ram and Vijay Shriram Sustainability Fellowship. In the past, I was a pre-doctoral fellow at the Energy Policy Institute (EPIC) at the University of Chicago, working at the Climate Impact Lab. In a previous life, I was a Data Scientist at DSaPP (now @CMU), and a research analyst at the Central Bank of Colombia.
research interests
I evaluate the effects of environmental changes on humans. I also develop machine learning models to create cool datasets that help us track humans and nature in data-scarce scenarios. I am interested in downscaling climate data products, multi-modal classification, measurement error in causal inference models, and wildfires in the Western US.
news
- Dec 2025 Orally presenting our Air Pollution Benefits paper at AGU Fall Meeting 2025 in New Orleans.
- Nov 2023 Accepted paper on wildfire house-burning risk at the CompSust workshop at NeurIPS 2023 in New Orleans.
- Sep 2023 Received the Stanford Data Science Fellowship (2 years) — now part of the SDS PhD Scholars 💻🤖.
- Aug 2023 Presenting our work on wildfire house-burning risk via multimodal classification and contrastive learning at TWEEDS in Portland.
publications
-
The air pollution benefits of low severity fire
Causally quantifies the air-quality benefits of prescribed (low-severity) fire relative to wildfire smoke, combining satellite measurements of fire behaviour with a synthetic-control design across the western US.
- paper
- EarthArXiv
- Code
-
Valuing wildfire smoke–related mortality benefits from climate mitigation
Estimates the avoided wildfire-smoke mortality and monetised health benefits associated with aggressive climate mitigation pathways in the contiguous United States.
- paper
- Paper
-
Effect of recent prescribed burning and land management on wildfire burn severity and smoke emissions in the western United States
Empirical analysis of how prescribed burning and recent land-management decisions have shaped wildfire burn severity and smoke emissions across the western US over the past two decades.
- paper
- Paper
-
A table is worth a thousand pictures: Multi-modal contrastive learning in house burning classification
CLIP-style multi-modal contrastive model that fuses NAIP aerial imagery with structured tabular features (rendered as text prompts) to classify whether individual houses burned during California wildfires (2010–2020).
- workshop paper
- OpenReview
-
Case study: Predictive fairness to reduce misdemeanor recidivism through social service interventions
Applies predictive-fairness audits to a social-service targeting model aimed at reducing misdemeanor recidivism. The accompanying tooling was released as the open-source
aequitaspackage.-
Protected Areas under Weak Institutions: Evidence from Colombia
Uses high-resolution remote-sensing measurements of deforestation in Colombia to test whether protected areas curb forest loss when state institutions are weak, finding heterogeneous effects that depend on the surrounding land-tenure regime.
- Ranked second-best paper by the International Sustainable Development Research Society.
- Press coverage (in Spanish): El Tiempo.
ongoing projects
-
A satellite foundation model for improved wealth monitoring
Pre-trained satellite-imagery foundation model whose embeddings improve poverty and wealth-asset prediction in data-scarce regions, particularly across Sub-Saharan Africa.
-
Arctic airmass displacement and reduced midlatitudes wintertime temperature variability under climate change
Investigates how Arctic warming displaces cold airmasses equator-ward and reduces wintertime temperature variability across the northern midlatitudes under climate change.
teaching
-
GEP 268: Topics and Methods in Global Environmental Policy I
Winter 2023 (as ESS 268), Winter 2025, Winter 2026
with Marshall Burke and Solomon Hsiang
TA’d Graduate-level course aimed at students who want to use applied econometrics to causally measure environmental change.
-
GEP 269: Topics and Methods in Global Environmental Policy II
Spring 2025, Spring 2026
with Marshall Burke and Solomon Hsiang
TA’d Graduate-level course aimed at students who want to use more advanced tools from causal inference and ML to measure environmental change. I led teaching sessions on balancing estimators and deep learning, and built and maintain the course website.